基於華語學習者語料庫之疑問詞「怎麼（樣）」偏誤分析

Corpus-based Analysis of Chinese Language Learners’ Errors in Using the “zenme(yang)” Wh-word

本研究旨在分析疑問詞句法語意特徵和學習者偏誤之關連性，作為教學規劃或評量設計時之參考指標。我們運用國家教育研究院中介語語料庫，分析台灣華語學習者疑問詞「怎麼（樣）」的使用情形。研究發現：一、句法位置與學習者偏誤有關聯。疑問詞出現在外狀語位置多誤加，內狀語位置多錯序，動詞組位置則多誤代；二、語意類型與學習者偏誤有關聯。表達驚嘆、原因、結果、程度義偏誤率高於平均水準，表達反詰、方法義偏誤率則低於平均水準；三、就母語背景和性別而言，越南學習者偏誤率較印尼學習者高，男性偏誤率高於女性；四、就語言類型而言，偏誤率最高是多重移位語言，最低則為疑問詞移位語言。建議教師可針對常見學習誤區，或不同母語背景學習者，設計課程，因材施教。

This study explored the correlation between the syntactic-semantic features of wh-words and Chinese language learners’ (CLLs’) errors. The results can be used as a reference for teaching planning or evaluation designing. The Chinese Learner Corpus was employed to analyze the use of the Chinese wh-word “zenme(yang).” We discovered the following: (1) the correlation between syntactic position and CLLs’ errors was significant. Outer-adverbial wh-words were often mistakenly added, inner-adverbial wh-words were often placed incorrectly within the sentence order, and verbal-phrase wh-words were often mistakenly replaced. (2) The association between semantic sentence type and CLLs’ errors was significant. The error rate for expressions including exclamations, reasons, results, and indicators of degree was higher than average, whereas that for expressions including rhetorical structures and indications of manner was lower than average. (3) With respect to language background and gender, the error rate of Vietnamese CLLs was 4.29%, whereas that of Indonesian CLLs was 27.59%, with the error rate of Indonesian CLLs being much higher than average. The error rate of men was higher than that of women, although the difference was nonsignificant. (4) Regarding language typology, CLLs whose mother tongues involving multiple wh-movement had the highest error rate (20%), whereas those speaking languages involving wh-movement had the lowest rate (3.64%). Teachers are recommended to teach students according to their proficiency and design courses by taking consideration into common learning difficulties or CLLs’ language backgrounds.